Purpose: Hypobaric hypoxic habitats are currently being touted as a potential solution to minimise decompression procedures in preparation for extra vehicular activities during future space missions. Since astronauts will live in hypoxic environments for the duration of such missions, the present study sought to elucidate the separate and combined effects of inactivity [simulated with the experimental bed rest (BR) model] and hypoxia on sleep characteristics in women.

Methods: Twelve women (Age = 27 ± 3 year) took part in three 10-day interventions, in a repeated measures cross-over counterbalanced design: (1) normobaric normoxic BR (NBR), (2) normobaric hypoxic BR (HBR; simulated altitude of 4,000 m), and (3) normobaric hypoxic ambulatory (HAMB; 4,000 m) confinement, during which sleep was assessed on night 1 and night 10 with polysomnography. In addition, one baseline sleep assessment was performed. This baseline assessment, although lacking a confinement aspect, was included statistically as a fourth comparison (i.e., pseudo normobaric normoxic ambulatory; pNAMB) in the present study.

Results: Hypoxia decreased sleep efficiency (p = 0.019), increased N1% sleep (p = 0.030), decreased N3 sleep duration (p = 0.003), and increased apnea hypopnea index (p < 0.001). BR impaired sleep maintenance, efficiency, and architecture [e.g., N2% sleep increased (p = 0.033)]. Specifically, for N3% sleep, the effects of partial pressure of oxygen and activity interacted. Hypoxia decreased N3% sleep both when active (pNAMB vs HAMB; p < 0.001) and inactive (NBR vs HBR; p = 0.021), however, this decrease was attenuated in the inactive state (-3.8%) compared to the active state (-10.2%).

Conclusion: A 10-day exposure to hypoxia and BR negatively impacted sleep on multiple levels as in macrostructure, microstructure and respiratory functioning. Interestingly, hypoxia appeared to have less adverse effects on sleep macrostructure while the participants were inactive (bed ridden) compared to when ambulatory. Data were missing to some extent (i.e., 20.8%). Therefore, multiple imputation was used, and our results should be considered as exploratory.